From: gnu
Date: Fri, 30 Sep 94 13:24:25 PDT
To: interesting-people@eff.org (interesting-people mailing list)
Subject: RC4
In-Reply-To: <aaa2f4be1e021004bc43@[130.91.88.102]>
Message-ID: <9409302023.AA21690@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


> SECRET FORMULA POSTED TO INTERNET
> The RSA encryption software RC4 owned by RSA Data Security, Inc., has been
> anonymously and illegally posted to electronic bulletin boards on the
> Internet, perhaps compromising the software's long-term effectiveness. (New
> York Times 9/17/94 p.17)

There was nothing obviously illegal about the posting, as far as I can
tell from the outside.  In private conversations over the years with
Jim Bidzos, President of RSA Data Security, he told me that RC4 was
held as a trade secret by RSA.  When I pressed him about why, rather
than patenting it and revealing the algorithm to the public, he said
it was a "business decision".

Revelation or publication of a trade secret is not illegal; trade
secrets are protected by contracts, not laws.  At worst, RSA has the
right to sue somebody who signed a contract with RSA, if such a
company disclosed RSA's source code.  But it's more likely that the
revelation was done by someone who never had a contract with RSA, by
reverse-engineering from widely available object code.  In that case,
RSA is unlikely to have a legal leg to stand on.

U.S. case law on reverse-engineering is spotty but tends to support
the right to examine copyrighted software in order to glean
uncopyrightable information (such as algorithms or interface
definitions) from it.  And there's no evidence that the reverse-
engineering even happened under U.S. law; most countries are more
permissive.  Since RC4 was deliberately marketed as an "exportable"
encryption algorithm, there are plenty of copies in countries all over
the world.

RSA would know whether the posted code's indentation style, block
structure, variable names, and lack of comments matched their own
source code, indicating that a source-code nondisclosure contract may
have been violated.  But they aren't saying, which probably means it
didn't match.  The lawyer-letter that RSA sent to the net was mere
bluster, similar to other threatening letters that RSA has sent over
the years.

The revelation of RC4 could help, or hurt, its long-term effectiveness.
RSA has always claimed that RC4 was secure if sufficiently long keys
are used, and its inventor, Ron Rivest, is well known for building
good ciphers.  Revelation may actually encourage the use of the
algorithm, if public scrutiny reveals its true strength.  This could
bring not only further fame to Ron Rivest, but also fortune to RSA,
which owns a fast, copyrighted implementation of RC4, and has plenty
of experience at selling cryptography to businesses.

RSA is not used to operating like an ordinary software publisher,
forced to actually compete with potential competitors rather than
clubbing them with lawsuits.  But it will have to learn that trick
soon anyway.  Its main patent will expire over the next decade, it
never had international rights anyway, and it's squabbling with
Cylink, its ex-partner in monopoly control of U.S. public key crypto.
I think the company has a potential to leverage the customer base and
cash flow from its patent into a strong competitive position in an
open market.

	John Gilmore